Deep statistical analysis and event forecasting for infrastructure

Where it will be useful

Network outsourcing, in-house infrastructure, startups.

See more →

How it works

Three steps: export logs → send the file → get a PDF report. From 10 minutes to 48 hours.

See more →

What the module finds

Eight analysis methods: priority, root causes, cycles, before/after, forecast, event links, MTBF/MTTR, industry norms.

See more →

Security

No access to your infrastructure. Isolated processing, automatic masking.

See more →

Audit levels

Free to try, then one-off audits — regular audits by agreement with the founder.

See more →

Run the analysis

Send an incident export - get a structured review. The first review is free.

Start →

Where it will be useful

Network outsourcing

Need a systematic way to work with clients?

Instead of long hours of manual review - a clear, structured report on all incidents, ready for the meeting.

In-house infrastructure

Stuck at your current SLA and looking for a way out?

Our analysis works as a second opinion: it shows the few problem groups behind 80% of downtime, and how to fix them.

Startup

Service quality cannot keep up with changes?

While you have no dedicated reliability engineer, get a structured picture of your risks without hiring one.

How it works

1

Export your logs

Any format works - whatever you already export from monitoring.

CSVJSONTXTXLSX
2

Send the file

Through the form on this page with your email, or straight to the Telegram bot. No signup, no install, no access to your infrastructure.

3

Get the PDF report

From 10 minutes to 48 hours, depending on the data: findings, risk points and recommendations, ready to show your team or client.

Example · Base

Full audit example - Base

6.5% of incidents = 80% of downtime, on real data. See the full PDF.

 Example · Granular

Full audit example - Granular

Plus tail risk, event chains and a cross-check by two models.

What our statistical module looks for

Eight core methods out of 20+ in the OpsLab engine. The full list also covers correlations, trend diagnostics, anomaly detection and a per-service breakdown.

Priority
Priority (Pareto ranking)
Which 20% of incidents cause 80% of the downtime, so you fix the few that matter. The Gini coefficient (a measure of inequality: 0 - all incidents equally bad, 1 - one incident caused everything) shows how concentrated the problem is.
3 of 90 incidents = 41% of all downtime. Gini = 0.67.
Root cause
Root causes (clustering)
Groups incidents by behaviour - time of day, duration, frequency - instead of by ticket category. A long list of "different" errors often turns out to be a small number of root causes.
90 incidents -> 2 clusters: "night, short" and "morning, long". Two causes, two teams.
Cycles
Cycles (periodicity)
Finds hidden repeat cycles in uneven time series - a method from astrophysics applied to incident logs. It also shows in which hours and days failures cluster. It reports FAP (false alarm probability - the chance the cycle is random noise).
Period = 23.1 h, FAP = 4x10⁻⁸ -> almost certainly a scheduler, not chance.
Verification
Before / after a change
Tests whether a deploy or a config change really moved the incident rate, and by how much. Cliff's delta (effect size - how big the shift is, from 0 to 1) turns "it feels worse" into a number.
Cliff's delta = +0.445, p < 0.001 -> the rate grew, confidence above 99.9%.
Forecast
Forecast and tail risk
Estimates how many failures to expect in the coming weeks, and the probability of a rare but very long outage - the one that breaks an SLA. The answer is always a range, never a falsely precise single number.
7 days - about 67 incidents (52-84). Chance of an outage longer than 6 hours - 7%.
Event links
Event links
Finds which events appear together and in what order, so an early alert can sit on the leading signal instead of the crash itself. Lift shows how many times more often a pair occurs than by chance.
database_timeout -> app_crash in 89% of cases, lift = 4.2.
Reliability
MTBF and MTTR without an agent
Mean time between failures (MTBF) and mean time to repair (MTTR) straight from your export - no agent, no access to your servers. Mean, median and P95 (the value only 5% of cases are worse than) together show the bad day, not only the typical one.
MTBF: 2.0 h mean, 0 h median, 24 h P95 - rare long pauses pull the mean up.
Benchmark
Industry norms
Puts your metrics next to the thresholds of your industry. The thresholds are data, not a hard-coded list. Where a threshold is unknown, the engine says so instead of inventing a norm.
P95 response 340 ms against a 200 ms telecom threshold; jitter 12 ms - within the 30 ms norm.
What comes next
Popular fixes after the analysis
Which problems we find most often and what to do about them, in ITIL / ITSM terms.

Run the analysis

CSV, JSON, TXT, XLSX - up to 20 MB.

📂
Drop your incident history file here
CSV · JSON · TXT · XLSX · up to 20 MB

Data can't leave your perimeter? Anonymize the file yourself first (free, offline script)

or
✈️
Telegram bot Send the file right in the chat - no email needed
→
✓ File received. The report will arrive at {email} from {from} in a few minutes. If you do not see it, check your spam folder.
Something went wrong. Please try again or send the file to the Telegram bot.

Security

✓

No access to your infrastructure

No agents, no VPN, no server access. Only the file you decided to send.

✓

Isolated processing

Every file is analysed in its own session. Client data never mixes.

✓

Deleted after 30 days, or sooner on request

Your file and the report page are visible only to you, on the site, for up to 30 days.

✓

No training on your data

Your data is never used to train AI models.

✓

Automatic masking

IP addresses, host names and emails are masked automatically before analysis.

✓

Anonymize before you send it

Run the offline script: it replaces service, host and people names with codes.

Download the script · How to run it

Audit levels

Order Base or Granular by email - we send an invoice, then deliver the paid report after payment. No card checkout yet.

Free $0
1 per week, by email
  • MTBF / MTTR
  • Periodicity - 1 finding
  • AI summary of the facts
  • 1-page PDF
  • No recommendations or charts
Base $200
one-off · per audit
  • Everything in Free
  • Pareto, clustering, periodicity
  • Before / after comparison
  • Forecasts and industry norms
  • AI review + top 3 recommendations
  • 3-page PDF with charts
Request Base →
Granular $600
one-off · per audit
  • Everything in Base
  • Tail risk (GPD)
  • Event links and lead-lag
  • Regime change point detection
  • Cross-check by two models
  • Extended PDF + appendix
  • Manual expert review, capped at 2 hours
Request Granular →

Need more than 2 hours of hands-on work? See contract diagnostics from $1500 →

Regular audits, by agreement
  • Audits on a schedule (weekly / monthly / per release)
  • A flexible set of methods
  • Terms based on volume and SLA
Discuss your incidents →